Glacial Erratics

These are some notes from IASummit 2006 held in lovely Vancouver, BC over the weekend of March 25th, 2006. I've cribbed these from my report back into the internal blog world of Socialtext who as my employer graciously packed me off to Vancouver to learn some stuff.    (Q05)

Notes and presentation    (Q06)

A growing list of notes and presentations to be found elsewhere.    (Q07)

• participant's notes of sessions I    (Q08)
• participant's notes of sessions II    (Q09)
• Jared Spool's We Are Not Alone: IA’s Role in Optimal Design Teams (I didn't go to this one but it was very well received.)    (Q0A)
• Peter Merholz Plenary Slides    (Q0B)
• Facets are Fundamental slides and notes    (Q0C)
• Information Architecture for the Spatial Web    (Q0D)

Note that my entire experience was somewhat clouded by the funk of hobbling around on my gimpy sprained ankle in one of the best walking cities around. Le sigh.    (Q0E)

This was the largest iasummit yet, with nearly 600 people, far more than expected, with over 50% first timers. It seemed that most were practitioners, followed by students, and then academics.    (Q0F)

Paraphrases of Note    (Q0G)

Things I heard in presentations or elsewhere that struck a cord. Callouts I called them in my palm pilot. No citations, sorry, that's just too hard to do in meatspace.    (Q0H)

• The internet world of information is becoming leaves in a big pile. The leaves have attributes that are applied to them from anywhere and everywhere.    (Q0I)
• Don't ever try to get consensus in enterprise space. You'll go down in flames.    (Q0J)
  • The implication of this one is that if you want to get anything done, it is best to move ahead. Especially in the enterprise space, if you wait around for everyone to agree, the project will die a horrible death.    (Q0K)
• Filtering in a system should occur on the way out, not on the way in.    (Q0L)
• Authority is a shortcut for learning/filtering. That is, we, humans, use the notion of authority so we can choose what to accept as accurate. Authority does not actually grant accuracy, it labels something as approved-by.    (Q0M)
• I'm pretty sure my Independent study that generated a paper on using associative structures to generate facets predicts a thing very much like tagging. The interface doesn't quite allow it but the idea was to let people create new descriptors. Tagging was all the rage at this year's summit. Facets and frameworks in years before. That paper was about all three.    (Q0N)
• If you've spent any time in the academic world of social informatics, this is old news but: Technology catalyzes awareness of existing tensions in the enterprise.    (Q0O)
  • The point the speaker was trying to make is that the role of the IA can often put you in the middle of the shit because trying to get things organized will upset a lot of people. The bigger point is that technology neither causes nor solves tensisions in the enterprise, it only exposes and assists.    (Q0P)
• Portals do not work.    (Q0Q)
  • This was said definitively and accepted as dogma by the audience.    (Q0R)
• Marking in content (in this case books) changes how people read.    (Q0S)
  • That is, when people using ebooks are given the opportunity to socially annotate the copy, people move around in the content much more, and reading becomes a much less linear activity.    (Q0T)
• Colocation (physical proximity) of people in the workspace provides opportunity for intuitive leaps that are unavailable to virtual teams. When colocation is not possible it is critical to be more explicit.    (Q0U)

Friday evening I enjoyed hanging with Jay Fienberg and meeting up with Roland Tanglao and Boris Mann from Bryght. Later met up with Christian Crumlish and Thomas Vander Wal    (Q0W)

Sushi was had. This is good. Lively discussion about information reuse, transclusion, etc.    (Q0X)

Notes from some of the presentations    (Q0Y)

Enterprise Information Architecture    (Q0Z)

A panel.    (Q10)

They gave some tips:    (Q11)

• Pay attention to search logs    (Q12)
• Always start with small bites when trying to improve information architecture and systems in the enterprise. Small successes lead more quickly to acceptance.    (Q13)
  • (A common theme throughout the conference: techniques for getting the job of IA accepted as necessary and valid for receiving resources.)    (Q14)
• Focus on specific tangible, measurable goal and tailor the benefit messages for each stakeholder audience.    (Q15)
• Always listen first prior to presenting your analysis of the situation.    (Q16)

The discussed obstacles to IA in the enterprise:    (Q17)

• Apathy    (Q18)
  • It's important to create a sense of urgency. Make the connection between bad IA and lost dollars. Execs at Motorola accept IA when you tell them: this is making it so you can't sell phones.    (Q19)
• Lack of unified vision    (Q1A)

Enterprise IA is primarily concerned with creating shared language (social and technical (i.e. taxonomies)) or methods for automating translation between different groups in the organization (again social and technical).    (Q1B)

One guy stressed that much of IA is to make up for poor leadership and management. To which I in my back row fantasy land stood up and spoke "Hallelujah Brother!"    (Q1C)

Interestingly my primary take away from this presentation was that information architecture, underneath the wireframe veneer is about business process analysis and consulting. I think all jobs in the US economy will trend in this direction over the next several years. That's really what a knowledge worker does: gets info on one end, puts info, designed to improve the situation out the other. I think this is kind of sad.    (Q1D)

Somewhere in here was some lunch, with most of the crowd from previous night's dinner. Drab but acceptable conference lunch fare. Afterwards I took some time to ice my foot.    (Q1E)

User Centered Design Considered Harmful    (Q1F)

The next presentation of note (there were others but I'm hitting the high points) was a review of Don Norman's Human-Centered Design Considered Harmful and some research done on reading e-books designed using activity centered design. This one struck me as quite relevant to the design of innovative collaborative applications.    (Q1G)

UCD is accused of limiting innovation because it attends to specific "tasks" that a "user" wants to "accomplish". ACD which builds on the back of situated action, distributed cognition and activity theory is more concerned with the larger environment, plans and goals of an individual or group. The example given had to do with books. A UCD approach to books might focus on the legibility of titles on bindings as it relates to the findability of a book in a library. An ACD approach is more concerned with the fact that the person wants to read something.    (Q1H)

This contrasts with my tool and task augmentation oriented approach.    (Q1I)

I think this is relevant in any environment that builds tools for customers because it sometimes feels like designers are not asking enough questions when a customer presents a feature request. Designers should try to unroll the request to get back to the activity the customers are trying to make better and work in concert with them to innovate new approaches to improving the activity, not just providing them with the feature.    (Q1J)

A process by any other name    (Q1K)

Presentation from Adam Polansky. Two things stood out:    (Q1L)

• Methodology does not matter. There is always, in this order:    (Q1M)
  • Idea    (Q1N)
  • Plan    (Q1O)
  • Build    (Q1P)
• The tradition: time, cost, quality pick 2 is wrong. In the real world it is actually pick 1    (Q1Q)

New approaches to Content Management    (Q1R)

Dan Brown presents his untested noodlings on a new way to do content managements systems. Throughout the conference there were a lot of rumblings about how much CMS suck: too many rules and too much structure getting in the way of changing environments. Dan's model sounded like a wiki to me, so I told him so afterwards, but since he was the hungriest guy in the room we adjourned for lunch rather than talking about it much.    (Q1S)

His general thought process was this: There is a metaphor that underlies CMS and it is wrong. Let's try on a different metaphor and see what this will do for us. The current metaphor, with us for a hundred or more years, is that business is a factory where there are products that are the result of a process. That's just not true. How about "business is a cell", with participants.    (Q1T)

The tasty tidbits from this presentation were twofold:    (Q1U)

• When generating content, rather than having prebuilt forms, make it possible to tag sections of the document for certain scenarios, which can then be used again for similar documents. The tags indicate what's a title, what's an author, if the document needs approval. This is tasty because it is attributes applied to stuff after the fact rather than requiring attributes at the start (see later comments about cognitive load of tagging).    (Q1V)
• When reviewing content, see who made contributions to a document using the rings of a tree as the interface to each revision. The width of the ring is the time the document was "owned" by that particular person, the color perhaps the amount of changes they made. Each ring can be moused over to gain access to revisions and other data, including RSS feeds for that particular author.    (Q1W)

Dan emphasized that his module highlights that people are making contributions, not fulfilling responsibilities. Rules exist in the system by process, not by computer enforcing rules. The computer helps with decisions rather than making them. That's some wikichurchspeak right there if I've ever heard it. And he didn't even know it.    (Q1X)

IA for Information Use and Reuse    (Q1Y)

Presentation by Thomas Vander Wal.    (Q1Z)

I was really excited about this presentation. It was the big standout in the program for me.    (Q20)

Something of a letdown. The IA community is months to years behind some aspects of the geek, protocol-oriented techno crowd. Thomas was essentially just saying the following phrase over and over again: "You've got to think about X" where X was microformats, use outside the web context, phones, people that don't live on the web.    (Q21)

I was hoping for something a little more noodly about what info reuse can mean for a community or society. It was essentially an expectation and context problem for me.    (Q22)

From Pace Layering to Resilience Theory    (Q23)

Best of Show. Karl Fast and Grant Campbell give a theory driven presentation discussing whether tagging is progressive or regressive. The IA community in general saw it as a step back last year, but is changing its tune this year. They announced their slides would be available but I can't find them yet.    (Q24)

Their answer: both.    (Q25)

Sorting, tagging and social information architecture    (Q26)

By Rashmi Sinha. Seemed like it was going to be good, a lot of cognitive analysis of sorting. Went a bit soft. Interesting bit was that tagging is often successful because it requires no sorting: you don't have to pick amongst categories to tag, you pick the concept or concepts that come to mind. Research has demonstrated that the cognitive load of typing a new tag is apparently less than picking a pre-defined category (I would call it class) to ascribe to an entity.    (Q27)

Closing Plenary    (Q28)

Peter Merholz gives everyone a group hug validating the existence of the profession.    (Q29)

This was followed by something called five minute madness. I was expecting something like YAPC lightning talks, but it was really just institutional confessionals and more group hugs. Group hugs in small groups are fine. With six hundred people I get a bit uncomfy. So I got out of there and headed for food before I got too snarky.    (Q2A)

And then the train home.    (Q2B)

Takeaways    (Q2C)

Besides the callouts above, things to remember or reflect upon:    (Q2D)

• Metadata for entitites that exists separately from the entities and out of the control of the entitities. No association other than saying so.    (Q2E)
• Pay more attention to general activities customers are trying to improve, less to their feature requests.    (Q2F)
• Adding one layer of semantics to tagging systems can be huge huge navigational win. For instance if 'cdent' is a tag that you can apply to whatever, by creating a small taxonomy of tags and saying that any tag can be 0 or 1 categories you can gain a faceted classification system: 'cdent' -> isa -> person. (This via Jay who went to Peter Van Dijck's presentaton.)    (Q2G)
• Tagging requires scale. Without scale it's a useful tool for findability for individuals to re-find their own stuff but of little important to other members of a group. We should think on this as we consider tagging in our app.    (Q2H)
• We need to help customers think about their information in terms of how it is or is not helping their bottom lines. We don't want to just help people work the way they want to work, we want that working to reap rewards.    (Q2I)

Thanks to Seb Paquet for giving me the push to publish this stuff.    (Q2J)

Hey wait, I need some techorati tags! iasummit iasummit2006 socialtext    (Q2K)